ABG-NAS: Adaptive bayesian genetic neural architecture search for graph representation learning

• We propose ABG-NAS, an adaptive NAS framework for graph representation learning. • A novel genetic search strategy dynamically balances exploration and exploitation. • Bayesian optimization is embedded to tune hyperparameters during the search process. • Our method outperforms SOTA GNAS methods on four benchmark graph datasets. • ABG-NAS achieves high F1 scores on both sparse and dense real-world graph structures. Effective and efficient graph representation learning is essential for enabling critical downstream tasks, such as node classification, link prediction, and subgraph search. However, existing graph neural network (GNN) architectures often struggle to adapt to diverse and complex graph structures, limiting their ability to produce structure-aware and task-discriminative representations. To address this challenge, we propose ABG-NAS, a novel framework for automated graph neural network architecture search tailored for efficient graph representation learning. ABG-NAS encompasses three key components: a Comprehensive Architecture Search Space (CASS), an Adaptive Genetic Optimization Strategy (AGOS), and a Bayesian-Guided Tuning Module (BGTM). CASS systematically explores diverse propagation ( P ) and transformation ( T ) operations, enabling the discovery of GNN architectures capable of capturing intricate graph characteristics. AGOS dynamically balances exploration and exploitation, ensuring search efficiency and preserving solution diversity. BGTM further optimizes hyperparameters periodically, enhancing the robustness and scalability of the resulting architectures to both large-scale graphs and high-complexity models. Empirical evaluations on benchmark datasets (Cora, PubMed, Citeseer, and CoraFull) demonstrate that ABG-NAS consistently outperforms both manually designed GNNs and state-of-the-art neural architecture search (NAS) methods. These results highlight the potential of ABG-NAS to advance graph representation learning by providing adaptive solutions that scale effectively across varying graph sizes and architectural complexities. Our code is publicly available at https://github.com/sserranw/ABG-NAS .